contrib/convert-repo

[PATCH] Generate correctly XML entities for obfuscated user

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

[PATCH] Generate correctly XML entities for obfuscated user

From: Edouard Gomez <[e-mail]>

manifest hash: 8e4e2d087ff60020c948d34e724fca99c84a9115
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)

iD8DBQFCxCpDywK+sNU5EO8RAs6yAKCy97yrMO6VYlkRIF3diLoGClZSOgCfekPE
ttPsLRoDTH12Tv6omFg6uUA=
=8ZBC
-----END PGP SIGNATURE-----

#!/usr/bin/env python
#
# This is a generalized framework for converting between SCM
# repository formats.
#
# In its current form, it's hardcoded to convert incrementally between
# git and Mercurial.
#
# To use, you must first import the first git version into Mercurial,
# and establish a mapping between the git commit hash and the hash in
# Mercurial for that version. This mapping is kept in a simple text
# file with lines like so:
#
# <git hash> <mercurial hash>
#
# To convert the rest of the repo, run:
#
# convert-repo <git-dir> <hg-dir> <mapfile>
#
# This updates the mapfile on each commit copied, so it can be
# interrupted and can be run repeatedly to copy new commits.

import sys, os, zlib, sha
from mercurial import hg, ui, util

class convert_git:
    def __init__(self, path):
        self.path = path

    def getheads(self):
        h = file(self.path + "/.git/HEAD").read()[:-1]
        return [h]

    def getfile(self, name, rev):
        a = file(self.path + ("/.git/objects/%s/%s"
                              % (rev[:2], rev[2:]))).read()
        b = zlib.decompress(a)
        if sha.sha(b).hexdigest() != rev: raise "bad hash"
        head, text = b.split('\0', 1)
        return text

    def getchanges(self, version):
        path = os.getcwd()
        os.chdir(self.path)
        fh = os.popen("git-diff-tree -m -r %s" % (version))
        os.chdir(path)
        
        changes = []
        for l in fh:
            if "\t" not in l: continue
            m, f = l[:-1].split("\t")
            m = m.split()
            h = m[3]
            p = (m[1] == "100755")
            changes.append((f, h, p))
        return changes

    def getcommit(self, version):
        c = self.getfile("", version) # read the commit hash
        end = c.find("\n\n")
        message = c[end+2:]
        l = c[:end].splitlines()
        manifest = l[0].split()[1]
        parents = []
        for e in l[1:]:
            n,v = e.split(" ", 1)
            if n == "author":
                p = v.split()
                date = " ".join(p[-2:])
                author = " ".join(p[:-2])
                if author[0] == "<": author = author[1:-1]
            if n == "committer": 
                p = v.split()
                date = " ".join(p[-2:])
                committer = " ".join(p[:-2])
                if committer[0] == "<": committer = committer[1:-1]
                message += "\ncommitter: %s %s\n" % (committer, date)
            if n == "parent": parents.append(v)
        return (parents, author, date, message)

class convert_mercurial:
    def __init__(self, path):
        self.path = path
        u = ui.ui()
        self.repo = hg.repository(u, path)

    def getheads(self):
        h = self.repo.changelog.heads()
        h = [ hg.hex(x) for x in h ]
        return h
        
    def putfile(self, f, e, data):
        self.repo.wfile(f, "w").write(data)
        util.set_exec(self.repo.wjoin(f), e)

    def delfile(self, f):
        try:
            os.unlink(self.repo.wjoin(f))
            self.repo.remove([f])
        except:
            pass

    def putcommit(self, files, parents, author, dest, text):
        if not parents: parents = ["0" * 40]
        if len(parents) < 2: parents.append("0" * 40)

        seen = {}
        pl = []
        for p in parents:
            if p not in seen:
                pl.append(p)
                seen[p] = 1
        parents = pl

        p2 = parents.pop(0)
        c = self.repo.changelog.count()
        while parents:
            p1 = p2
            p2 = parents.pop(0)
            self.repo.rawcommit(files, text, author, dest, 
                                hg.bin(p1), hg.bin(p2))
            text = "(octopus merge fixup)\n"

        return hg.hex(self.repo.changelog.node(c))

class convert:
    def __init__(self, source, dest, mapfile):
        self.source = source
        self.dest = dest
        self.mapfile = mapfile
        self.commitcache = {}

        self.map = {}
        for l in file(self.mapfile):
            sv, dv = l[:-1].split()
            self.map[sv] = dv

    def walktree(self, heads):
        visit = heads
        known = {}
        parents = {}
        while visit:
            n = visit.pop(0)
            if n in known or n in self.map: continue
            known[n] = 1
            self.commitcache[n] = self.source.getcommit(n)
            cp = self.commitcache[n][0]
            for p in cp:
                parents.setdefault(n, []).append(p)
                visit.append(p)

        return parents

    def toposort(self, parents):
        visit = parents.keys()
        seen = {}
        children = {}
        while visit:
            n = visit.pop(0)
            if n in seen: continue
            seen[n] = 1
            pc = 0
            if n in parents:
                for p in parents[n]:
                    if p not in self.map: pc += 1
                    visit.append(p)
                    children.setdefault(p, []).append(n)
            if not pc: root = n

        s = []
        removed = {}
        visit = parents.keys()
        while visit:
            n = visit.pop(0)
            if n in removed: continue
            dep = 0
            if n in parents:
                for p in parents[n]:
                    if p in self.map: continue
                    if p not in removed:
                        # we're still dependent
                        visit.append(n)
                        dep = 1
                        break

            if not dep:
                # all n's parents are in the list
                removed[n] = 1
                s.append(n)
                if n in children:
                    for c in children[n]:
                        visit.insert(0, c)

        return s

    def copy(self, rev):
        p, a, d, t = self.commitcache[rev]
        files = self.source.getchanges(rev)

        for f,v,e in files:
            try:
                data = self.source.getfile(f, v)
            except IOError, inst:
                self.dest.delfile(f)
            else:
                self.dest.putfile(f, e, data)

        r = [self.map[v] for v in p]
        f = [f for f,v,e in files]
        self.map[rev] = self.dest.putcommit(f, r, a, d, t)
        file(self.mapfile, "a").write("%s %s\n" % (rev, self.map[rev]))

    def convert(self):
        heads = self.source.getheads()
        parents = self.walktree(heads)
        t = self.toposort(parents)
        num = len(t)

        for c in t:
            num -= 1
            if c in self.map: continue
            desc = self.commitcache[c][3].splitlines()[0]
            print num, desc
            self.copy(c)

gitpath, hgpath, mapfile = sys.argv[1:]

c = convert(convert_git(gitpath), convert_mercurial(hgpath), mapfile)
c.convert()